"Amnesia" - A Selection of Machine Learning Models That Can Forget User Data Very Fast
Summary: Amnesia: efficient decremental update algorithms that remove a user’s contribution from trained ML models without re-accessing original training data, enabling fast compliance with right-to-be-forgotten. Rust implementations for four common ML methods show orders-of-magnitude speedups on nine real datasets and outline Differential Dataflow parallelization plus limitations. (summarized by gpt-5-mini on Feb 09 2026)
Incoming Non-self Citations Over Time
Authors
- 1. Sebastian Schelter (New York University)
BibTeX Citation
@inproceedings{schelter_cidr20,
address = {Amsterdam, Netherlands},
series = {{CIDR} '20},
title = {{"Amnesia" - A Selection of Machine Learning Models That Can Forget User Data Very Fast}},
booktitle = {Proceedings of the {Conference} on {Innovative} {Data} {Systems} {Research}},
author = {Schelter, Sebastian},
year = {2020}
}Incoming Citations (Sorted by Pagerank)
Showing 6 of 6 citing papers.
| Rank | Citing Paper | Year | Venue | Pagerank |
|---|---|---|---|---|
| 3,960 | HedgeCut: Maintaining Randomised Trees for Low-Latency Machine Unlearning | 2021 | SIGMOD | 6.9878154e-05 |
| 4,240 | LIMA: Fine-grained Lineage Tracing and Reuse in Machine Learning Systems | 2021 | SIGMOD | 6.809685e-05 |
| 7,381 | DeltaBoost: Gradient Boosting Decision Trees with Efficient Machine Unlearning | 2023 | SIGMOD | 5.6288368e-05 |
| 7,687 | ExDRa: Exploratory Data Science on Federated Raw Data | 2021 | SIGMOD | 5.5671645e-05 |
| 11,302 | Snapcase – Regain Control over Your Predictions with Low-Latency Machine Unlearning | 2024 | VLDB | 5.093636e-05 |
| 11,509 | Screening Native ML Pipelines with “ArgusEyes” | 2022 | CIDR | 5.093636e-05 |
Previous
Page 1 / 1
Next
Outgoing Citations (Sorted by Pagerank)
Showing 3 of 3 cited papers.
Citations counted here include only citations to other VLDB/SIGMOD/CIDR/PODS papers in this database.
| Rank | Cited Paper | Year | Venue | Pagerank |
|---|---|---|---|---|
| 21 | Similarity Search in High Dimensions via Hashing | 1999 | VLDB | 0.00056760516 |
| 455 | Differential dataflow | 2013 | CIDR | 0.00018133241 |
| 2,196 | Spinning Fast Iterative Data Flows | 2012 | VLDB | 8.9704984e-05 |
Previous
Page 1 / 1
Next
Semantically Similar Papers
| # | Overall Rank | Paper | Year | Venue |
|---|---|---|---|---|
| 1 | 11,063 | New Trends in Data Forgetting for Sustainable Data Management | 2025 | VLDB |
| 2 | 10,925 | Meaningful Data Erasure in the Presence of Dependencies | 2025 | VLDB |
| 3 | 3,960 | HedgeCut: Maintaining Randomised Trees for Low-Latency Machine Unlearning | 2021 | SIGMOD |
| 4 | 11,302 | Snapcase – Regain Control over Your Predictions with Low-Latency Machine Unlearning | 2024 | VLDB |
| 5 | 8,908 | Privacy and Accuracy-Aware AI/ML Model Deduplication | 2025 | SIGMOD |
| 6 | 10,404 | Stochastic Submodular Data Forgetting | 2026 | SIGMOD |
| 7 | 5,820 | A Database System with Amnesia | 2017 | CIDR |
| 8 | 7,381 | DeltaBoost: Gradient Boosting Decision Trees with Efficient Machine Unlearning | 2023 | SIGMOD |
| 9 | 9,587 | Machine Unlearning in Learned Databases: An Experimental Analysis | 2024 | SIGMOD |
| 10 | 7,328 | On the Feasibility of Forgetting in Data Streams | 2024 | PODS |